RESOLVED FIXED 92012
IndexedDB: Temporarily continue generating backend keys for empty key lists 
https://bugs.webkit.org/show_bug.cgi?id=92012
Summary IndexedDB: Temporarily continue generating backend keys for empty key lists
Alec Flett
Reported 2012-07-23 11:30:09 PDT
IndexedDB: Temporarily continue generating backend keys for empty key lists
Attachments
Patch (2.49 KB, patch)
2012-07-23 11:33 PDT, Alec Flett 		no flags
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
Alec Flett
Comment 1 2012-07-23 11:33:17 PDT
Created attachment 153822 [details] Patch
Alec Flett
Comment 2 2012-07-23 11:44:07 PDT
I was trying to land the chromium side of things, which I thought worked, until I realized it was the chromium browser tests of idb that were failing. the problem I was having was that to move from put() to putWithIndexKeys(), I need to pass NULL rather than an empty array to trigger the old (backend) key generation - but from inside the proxy, there isn't really a way to pass 'null' - which meant that all the calls through the chromium proxy code were telling the webkit code not to generate any keys. I tried to fix this on the chromium side, but it was going to involve a whole complicated dance adding a further boolean 'generateIndexKeys' boolean to the IPC, picking it up again on the chromium side, etc. The somewhat hacky, but definitely temporary fix is to just always try to generate keys on the backend if there are no index keys passed. This won't actually cause any problems - previously this meant "for the zero indexes you need to fill, here are the values" and then it would go generate keys for 0 indexes, essentially a no-op. Now, it means "there are no key index values, so go generate them" - this means we will briefly generate duplicate index keys while I land both sides of this, but in terms of correctness, there is no change. As soon as *this* patch lands and webkit rolls, I can check the "commit" checkbox in http://codereview.chromium.org/10786043/
Alec Flett
Comment 3 2012-07-23 11:44:42 PDT
jsbell@ mind taking a quick look before I get a r+?
Joshua Bell
Comment 4 2012-07-23 16:40:22 PDT
Comment on attachment 153822 [details] Patch lgtm
Alec Flett
Comment 5 2012-07-24 09:36:17 PDT
tony@ - a quick one. r?
WebKit Review Bot
Comment 6 2012-07-24 10:11:36 PDT
Comment on attachment 153822 [details] Patch Clearing flags on attachment: 153822 Committed r123489: <http://trac.webkit.org/changeset/123489>
WebKit Review Bot
Comment 7 2012-07-24 10:11:39 PDT
All reviewed patches have been landed. Closing bug.
Note You need to log in before you can comment on or make changes to this bug.